model-api-mapping.json

{
  "title": "Model-to-API Endpoint Mapping",
  "description": "Maps AI model families to their supported API endpoints. Used to determine which endpoint (chat/completions vs responses vs messages) a model requires.",
  "lastUpdated": "2026-07-11T06:00:54Z",
  "sources": {
    "openai": "https://platform.openai.com/docs/models",
    "anthropic": "https://docs.anthropic.com/en/docs/about-claude/models"
  },
  "providers": {
    "openai": {
      "endpoints": {
        "responses": {
          "path": "/v1/responses",
          "description": "Newer unified API for agentic, tool-using, and multimodal workflows. Required for GPT-5+ models; some older models (o1, o3, gpt-4o, gpt-4.1) support it alongside chat/completions."
        },
        "chat_completions": {
          "path": "/v1/chat/completions",
          "description": "Legacy conversational API. Being deprecated for newer models by end of 2026."
        }
      },
      "models": [
        {
          "family": "gpt-5.6",
          "patterns": ["gpt-5.6*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Supports both endpoints. Includes gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna per OpenAI SDK ChatModel."
        },
        {
          "family": "gpt-5.5",
          "patterns": ["gpt-5.5*"],
          "endpoints": ["responses"],
          "notes": "Responses API only. Not present in OpenAI OpenAPI spec v2.3.0; reserved for future gpt-5.5 family models."
        },
        {
          "family": "gpt-5.4",
          "patterns": ["gpt-5.4*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Supports both endpoints on OpenAI. Includes gpt-5.4-mini, gpt-5.4-nano. NOTE: gpt-5.4-mini is not accessible via /chat/completions on the Copilot API (responses-only via Copilot); the api-proxy handles this by falling back to the next alias candidate."
        },
        {
          "family": "gpt-5.3",
          "patterns": ["gpt-5.3*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Supports both endpoints. Only gpt-5.3-chat-latest confirmed in OpenAPI spec; gpt-5.3-codex variants not present."
        },
        {
          "family": "gpt-5.2",
          "patterns": ["gpt-5.2*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Supports both endpoints. Includes gpt-5.2-pro."
        },
        {
          "family": "gpt-5.1-codex-max",
          "patterns": ["gpt-5.1-codex-max"],
          "endpoints": ["responses"],
          "notes": "Responses API only (ResponsesOnlyModel)."
        },
        {
          "family": "gpt-5.1",
          "patterns": ["gpt-5.1*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Supports both endpoints for gpt-5.1, gpt-5.1-codex, gpt-5.1-mini."
        },
        {
          "family": "gpt-5-codex/pro",
          "patterns": ["gpt-5-codex", "gpt-5-pro", "gpt-5-pro-*"],
          "endpoints": ["responses"],
          "notes": "Responses API only for gpt-5-codex and gpt-5-pro (ResponsesOnlyModel). Includes dated variants like gpt-5-pro-2025-10-06."
        },
        {
          "family": "gpt-5",
          "patterns": ["gpt-5", "gpt-5-*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Supports both endpoints for gpt-5, gpt-5-mini, gpt-5-nano."
        },
        {
          "family": "o4-mini-deep-research",
          "patterns": ["o4-mini-deep-research", "o4-mini-deep-research-*"],
          "endpoints": ["responses"],
          "notes": "Responses API only (ResponsesOnlyModel). Includes dated variants like o4-mini-deep-research-2025-06-26."
        },
        {
          "family": "o4",
          "patterns": ["o4*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Reasoning model. o4-mini supports both endpoints (ModelIdsShared)."
        },
        {
          "family": "o3-pro/deep-research",
          "patterns": ["o3-pro", "o3-pro-*", "o3-deep-research", "o3-deep-research-*"],
          "endpoints": ["responses"],
          "notes": "Responses API only. o3-pro and o3-deep-research are in ResponsesModel but not ChatModel."
        },
        {
          "family": "o3",
          "patterns": ["o3*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Reasoning model. o3, o3-mini support both endpoints. o3-pro/deep-research use responses only (covered by separate entry above)."
        },
        {
          "family": "o1-pro",
          "patterns": ["o1-pro", "o1-pro-*"],
          "endpoints": ["responses"],
          "notes": "Responses API only. o1-pro is in ResponsesModel but not ChatModel."
        },
        {
          "family": "o1",
          "patterns": ["o1*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Reasoning model. o1, o1-mini, o1-preview support both endpoints. o1-pro uses responses only (covered by separate entry above)."
        },
        {
          "family": "computer-use-preview",
          "patterns": ["computer-use-preview*"],
          "endpoints": ["responses"],
          "notes": "Responses API only. Specialized model for computer use tasks."
        },
        {
          "family": "codex-mini",
          "patterns": ["codex-mini*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Supports both endpoints. codex-mini-latest is in ChatModel."
        },
        {
          "family": "chatgpt-4o-latest",
          "patterns": ["chatgpt-4o-latest"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Rolling alias for the latest ChatGPT-optimized GPT-4o model. Supports both endpoints (ModelIdsShared). Not matched by gpt-4o* pattern due to chatgpt- prefix."
        },
        {
          "family": "gpt-4o",
          "patterns": ["gpt-4o*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Multimodal. Supports both endpoints. Includes audio and search preview variants."
        },
        {
          "family": "gpt-4.1",
          "patterns": ["gpt-4.1*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Supports both endpoints. Includes gpt-4.1-mini, gpt-4.1-nano."
        },
        {
          "family": "gpt-4.5",
          "patterns": ["gpt-4.5*"],
          "endpoints": ["chat_completions"],
          "notes": "Preview model. Chat completions only."
        },
        {
          "family": "gpt-4-turbo",
          "patterns": ["gpt-4-turbo*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Legacy. In ModelIdsShared per OpenAI spec v2.3.0; supports both endpoints."
        },
        {
          "family": "gpt-4",
          "patterns": ["gpt-4", "gpt-4-*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Legacy. In ModelIdsShared per OpenAI spec v2.3.0; supports both endpoints."
        },
        {
          "family": "gpt-3.5-turbo",
          "patterns": ["gpt-3.5-turbo*"],
          "endpoints": ["chat_completions", "responses"],
          "notes": "Legacy. In ModelIdsShared per OpenAI spec v2.3.0; supports both endpoints. Deprecated."
        }
      ]
    },
    "anthropic": {
      "endpoints": {
        "messages": {
          "path": "/v1/messages",
          "description": "Unified Messages API for all Claude models. Supports tool use, streaming, extended thinking, and vision."
        }
      },
      "models": [
        {
          "family": "claude-fable-5",
          "patterns": ["claude-fable-5*"],
          "endpoints": ["messages"],
          "notes": "Latest flagship. 1M token context."
        },
        {
          "family": "claude-mythos-5",
          "patterns": ["claude-mythos-5*", "claude-mythos-preview*"],
          "endpoints": ["messages"],
          "notes": "Invitation-only (Project Glasswing). Limited availability. 1M token context."
        },
        {
          "family": "claude-sonnet-5",
          "patterns": ["claude-sonnet-5*"],
          "endpoints": ["messages"],
          "notes": "Best combination of speed and intelligence. 1M token context."
        },
        {
          "family": "claude-opus-4",
          "patterns": ["claude-opus-4*"],
          "endpoints": ["messages"],
          "notes": "Includes claude-opus-4-8, claude-opus-4-7, claude-opus-4-6, claude-opus-4-5, claude-opus-4-1. 1M token context."
        },
        {
          "family": "claude-sonnet-4",
          "patterns": ["claude-sonnet-4*"],
          "endpoints": ["messages"],
          "notes": "Includes claude-sonnet-4-6, claude-sonnet-4-5. Production workhorse. 1M token context."
        },
        {
          "family": "claude-haiku-4",
          "patterns": ["claude-haiku-4*"],
          "endpoints": ["messages"],
          "notes": "Includes claude-haiku-4-5. Fast/cheap. 200K token context."
        },
        {
          "family": "claude-3.5-sonnet",
          "patterns": ["claude-3-5-sonnet*", "claude-3.5-sonnet*"],
          "endpoints": ["messages"],
          "notes": "Previous generation. 200K token context."
        },
        {
          "family": "claude-3.5-haiku",
          "patterns": ["claude-3-5-haiku*", "claude-3.5-haiku*"],
          "endpoints": ["messages"],
          "notes": "Previous generation fast model."
        },
        {
          "family": "claude-3-opus",
          "patterns": ["claude-3-opus*"],
          "endpoints": ["messages"],
          "notes": "Previous generation. 200K token context."
        }
      ]
    }
  }
}